Modeling of the CPS1 for the Frequency Constrained Unit Commitment

Frequency is an important index to reflect stability of the power system. NERC's Control Performance Standard 1 (CPSI) is to assess control area's ability in maintaining frequency quality. This paper proposes a stepwise regression method to identify important frequency-control correlation factors for developing a CPSI model. The CPSI model is constructed by the Long Short-Term Memory (LSTM) neural network. The well trained CPSI model is then integrated into a unit commitment (UC) program to realize a CPSI compliant UC (UC-CPSI). The performance of the proposed UC-CPSI is validated by simulation case and is compared with the case without the frequency constrained UC.